ADC Input Noise: The Good,

The Bad, and The Ugly.

Is No Noise Good Noise?
By Walt Kester [walt.kester@analog.com]
Introduction

All analog-to-digital converters (ADCs) have a certain amount of

input-referred noisemodeled as a noise source connected in series
with the input of a noise-free ADC. Input-referred noise is not
to be confused with quantization noise, which is only of interest
when an ADC is processing time-varying signals. In most cases,
less input noise is better; however, there are some instances where
input noise can actually be helpful in achieving higher resolution.
If this doesnt seem to make sense right now, read on to find out
how some noise can be good noise.

input rms noise. See Further Reading 6 for a detailed description

of how to calculate the value of from the histogram data. It is
common practice to express this rms noise in terms of LSBs
rms, corresponding to an rms voltage referenced to the ADC
full-scale input range. If the analog input range is expressed as
digital numbers, or counts, input values, such as , can be expressed
as a count of the number of LSBs.
P-P INPUT NOISE
6.6 RMS NOISE

NUMBER OF
OCCURRENCES

STANDARD DEVIATION
= RMS NOISE (LSBs)

Input-Referred Noise (Code-Transition Noise)

Practical ADCs deviate from ideal ADCs in many ways. Inputreferred noise is certainly a departure from the ideal, and its
effect on the overall ADC transfer function is shown in Figure 1.
As the analog input voltage is increased, the ideal ADC (shown
in Figure 1a) maintains a constant output code until a transition
region is reached, at which point it instantly jumps to the next
value, remaining there until the next transition region is reached.
A theoretically perfect ADC has zero code-transition noise, and a
transition region width equal to zero. A practical ADC has a
certain amount of code transition noise, and therefore a finite
transition region width. Figure 1b shows a situation where
the width of the code transition noise is approximately one
least-significant bit (LSB) peak-to-peak.
(a) IDEAL ADC
DIGITAL
OUTPUT

(b) ACTUAL ADC

DIGITAL
OUTPUT

ANALOG
INPUT

n4 n3 n2 n1
n
n+1 n+2 n+3 n+4
OUTPUT CODE

Figure 2. Effect of input-referred noise on ADC groundedinput histogram for an ADC with a small amount of DNL.
Although the inherent differential nonlinearity (DNL) of the ADC
will cause deviations from an ideal Gaussian distribution (for
instance, some DNL is evident in Figure 2), it should be at least
approximately Gaussian. If there is significant DNL, the value
of should be calculated for several different dc input voltages
and the results averaged. If the code distribution is significantly
non-Gaussian, as exemplified by large and distinct peaks and
valleys, for instancethis could indicate either a poorly designed
ADC ormore likelya bad PC board layout, poor grounding
techniques, or improper power supply decoupling (see Figure 3).
Another indication of trouble is when the width of the distribution
changes drastically as the dc input is swept over the ADC input
voltage range.
NUMBER OF
OCCURRENCES

ANALOG
INPUT

Figure 1. Code-transition noise (input-referred noise)

and its effect on ADC transfer function.
Internally, all ADC circuits produce a certain amount of rms noise
due to resistor noise and kT/C noise. This noise, present even
for dc input signals, accounts for the code-transition noise, now
generally referred to as input-referred noise. Input-referred noise is
most often characterized by examining the histogram of a number
of output samples, while the input to the ADC is held constant at a
dc value. The output of most high speed or high resolution ADCs
is a distribution of codes, typically centered around the nominal
value of the dc input (see Figure 2).
To measure the amount of input-referred noise, the input of the
ADC is either grounded or connected to a heavily decoupled
voltage source, and a large number of output samples are
collected and plotted as a histogram (referred to as a groundedinput histogram if the input is nominally at zero volts). Since the
noise is approximately Gaussian, the standard deviation of the
histogram, , which can be calculated, corresponds to the effective

Analog Dialogue 40-02, February (2006)

n5 n4 n3 n2 n1

n+1 n+2 n+3 n+4 n+5

OUTPUT CODE

Figure 3. Grounded-input histogram for poorly designed

ADC and/or poor layout, grounding, or decoupling.

Noise-Free (Flicker-Free) Code Resolution

The noise-free code resolution of an ADC is the number of bits

of resolution beyond which it is impossible to distinctly resolve
individual codes. This limitation is due to the effective input noise
(or input-referred noise) associated with all ADCs and described
above, usually expressed as an rms quantity with the units of
LSBs rms. Multiplying by a factor of 6.6 converts the rms noise into
a useful measure of peak-to-peak noisethe actual uncertainty with
which a code can be identifiedexpressed in LSBs peak-to-peak.

http://www.analog.com/analogdialogue

Peak-to-Peak Resolution vs. Input Range and Update Rate (CHP = 1)

Peak-to-Peak Resolution in Counts (Bits)
Output
Data Rate
50 Hz
100 Hz
150 Hz
200 Hz*
400 Hz

3 dB
Frequency
1.97 Hz
3.95 Hz
5.92 Hz
7.9 Hz
15.8 Hz

SF
Word
2048
1024
683
512
256

Settling Time
Normal Mode
460 ms
230 ms
153 ms
115 ms
57.5 ms

Settling Time
Fast Mode
60 ms
30 ms
20 ms
15 ms
7.5 ms

Input Range
= 80 mV
230k (18)
170k (17.5)
130k (17)
120k (17)
80k (16.5)

Input Range
= 40 mV
175k (17.5)
125k (17)
100k (16.5)
90k (16.5)
55k (16)

Input Range
= 20 mV
120k (17)
90k (16.5)
70k (16)
65k (16)
40k (15.5)

Input Range
= 10 mV
80k (16.5)
55k (16)
45k (15.5)
40k (15.5)
30k (15)

*Power-On Default

Figure 4. Noise-free code resolution for the AD7730 sigma-delta ADC.

Since the total range (or span) of an N-bit ADC is 2N LSBs, the
total number of noise-free counts is therefore equal to:

Noise-free counts =

2N
peak -to-peak input noise ( LSBs ) (1)

The number of noise-free counts can be converted into noise-free

(binary) code resolution by calculating the base-2 logarithm as
follows:

2N
Noise-free code resolution = log 2
(2)
peak -to-peak input noise ( LSBs )

The noise-free code resolution specification is generally associated

with high-resolution sigma-delta measurement ADCs. It is most
often a function of sampling rate, digital-filter bandwidth, and
programmable-gain-amplifier (PGA) gain (hence input range).
Figure 4 shows a typical tabletaken from the data sheet of the
AD7730 sigma-delta ADC1.
Note that for an output data rate of 50 Hz and an input range
of 610 mV, the noise-free code resolution is 16.5 bits (80,000
noise-free counts). The settling time under these conditions is
460 ms, making this ADC an ideal candidate for a precision
weigh-scale application. Data of this kind is available on most data
sheets for high-resolution sigma-delta ADCs suitable for precision
measurement applications.
The ratio of the full-scale range to the rms input noise (rather than
peak-to-peak noise) is sometimes used to calculate resolution.
In this case, the term effective resolution is used. Note that under
identical conditions, effective resolution is larger than noise-free code
resolution by log2 (6.6), or approximately 2.7 bits.

2N
Effective resolution = log 2

rms
input
noise
(
LSBs
)

Effective resolution = Noise-free code resolution + 2.7 bits (4)

(3)

Some manufacturers prefer to specify effective resolution rather

than noise-free code resolution because it results in a higher
number of bitsthe user should check the data sheet closely to
make sure which is actually specified.

Digital Averaging Increases Resolution and Reduces Noise

The effects of input-referred noise can be reduced by digital

averaging. Consider a 16-bit ADC which has 15 noise-free bits
at a sampling rate of 100 kSPS. Averaging two measurements
of an unchanging signal for each output sample reduces the
effective sampling rate to 50 kSPSand increases the SNR by
3 dB and the number of noise-free bits to 15.5. Averaging four
measurements per output sample reduces the sampling rate to
25 kSPSand increases the SNR by 6 dB and the number of
noise-free bits to 16.

We can go even further and average 16 measurements per output;

the output sampling rate is reduced to 6.25 kSPS, the SNR
increases by another 6 dB, and the number of noise-free bits
increases to 17. The arithmetic precision in the averaging must
be carried out to the larger number of significant bits in order to
gain the extra resolution.
The averaging process also helps smooth out the DNL errors
in the ADC transfer function. This can be illustrated for the
simple case where the ADC has a missing code at quantization
level k. Even though code k is missing because of the large DNL
error, the average of the two adjacent codes, k 1 and k + 1, is
equal to k.
This technique can therefore be used effectively to increase
the dynamic range of the ADC at the expense of overall output
sampling rate and extra digital hardware. It should also be noted
that averaging will not correct the inherent integral nonlinearity
of the ADC.
Now, consider the case of an ADC that has extremely low
input-referred noise, and the histogram shows a single code
no matter how many samples are taken. What will digital
averaging do for this ADC? This answer is simpleit will
do nothing! No matter how many samples are averaged, the
answer will be the same. However, as soon as enough noise is
added to the input signal, so that there is more than one code
in the histogram, the averaging method starts working again.
Thusinterestinglysome small amount of noise is good (at
least with respect to the averaging method); however, the more
noise present at the input, the more averaging is required to
achieve the same resolution.

Dont Confuse Effective Number of Bits (ENOB) with Effective

Resolution or Noise-Free Code Resolution

Because of the similarity of the terms, effective number of bits

and effective resolution are often assumed to be equal. This is
not the case.
Effective number of bits (ENOB) is derived from an FFT
analysis of the ADC output when the ADC is stimulated with
a full-scale sine-wave input signal. The root-sum-of-squares
(RSS) value of all noise and distortion terms is computed, and
the ratio of the signal to the noise-and-distortion is defined as
SINAD, or S/(N+D). The theoretical SNR of a perfect N-bit
ADC is given by:

SNR = 6.02 N + 1.76 dB

(5)

ENOB is calculated by substituting the ADCs computed

SINAD for SNR in Equation 5 and solving equation for N.

ENOB =

SINAD 1.76 dB
6.02

(6)

Analog Dialogue 40-02, February (2006)

The noise and distortion used to calculate SINAD and ENOB

include not only the input-referred noise but also the quantization
noise and the distortion terms. SINAD and ENOB are used to
measure the dynamic performance of an ADC, while effective
resolution and noise-free code resolution are used to measure the
noise of the ADC under essentially dc input conditions, where
quantization noise is not an issue.

this scheme improves distortion produced by the ADCs encoder

nonlinearity, it does not significantly improve distortion created
by its front end.
(a)
SMALL AMPLITUDE
INPUT
+
+

Using Noise Dither to Increase an ADCs Spurious-Free

Dynamic Range

Another method, one that is easier to implementespecially

in wideband receiversis to inject a narrow-band dither signal
outside the signal band of interest, as shown in Figure 6. Usually,
no signal components are located in the frequency range near
dc, so this low-frequency region is often used for such a dither
signal. Another possible location for the dither signal is slightly
below f S /2. The dither signal occupies only a small bandwidth
relative to the signal bandwidth (usually a bandwidth of a few
hundred kHz is sufficient), so no significant degradation in SNR
occursas it would if the dither was broadband.
fS
INPUT

NOISE
GENERATOR

OUT-OF-BAND
FILTER

OUT-OF-BAND NOISE
NEAR DC OR fS/2

Figure 6. Injecting out-of-band dither to improve ADC SFDR.

A subranging, pipelined ADC, such as the AD6645 14-bit,
105-MSPS ADC2 (see Figure 7), has very small differential
nonlinearity errors that occur at specific code transition points
across the ADC range. The AD6645 includes a 5-bit ADC
(ADC1), followed by a 5-bit ADC2 and a 6-bit ADC3. The only
significant DNL errors occur at the ADC1 transition pointsthe
second- and third-stage DNL errors are minimal. There are
25 = 32 decision points associated with ADC1, which occur every
68.75 mV (29 = 512 LSBs) for a 2.2-V full-scale input range.

DVCC

A1

TH1

TH2

A2

ADC1

2.4V

AD6645
TH3

TH4

DAC1

DAC2
5

INTERNAL
TIMING

GND

ADC3

TH5

ADC2

5
ENCODE

ADC

BPF

25 = 32 ADC1 TRANSITIONS

ENCODE

RANDOM
NUMBER
GENERATOR

Figure 5. Using dither to randomize ADC transfer function.

Other schemes have been developed using larger amounts of dither

noise to randomize the transfer function of the ADC. Figure 5b
shows a dither-noise source comprising a pseudo-random-number
generator driving a DAC. This signal is subtracted from the
ADC input signal and then digitally added to the ADC output,
thereby causing no significant degradation SNR. An inherent
disadvantage of this technique, however, is that the input signal
swing must be reduced to prevent overdriving the ADC as the
amplitude of the dither signal is increased. Note that although

VREF

ADDER

DAC

Dithering can be used to improve SFDR of an ADC under certain

conditions (see Further Reading 25). For example, even in a
perfect ADC, some correlation exists between the quantization
noise and the input signal. This correlation can reduce the SFDR
of the ADC, especially if the input signal is an exact sub-multiple
of the sampling frequency. Summing about 1/ 2 -LSB rms of
broadband noise with the input signal tends to randomize the
quantization noise and minimize this effect (see Figure 5a). In
most systems, however, the noise already riding on top of the signal
(including the input-referred noise of the ADC) obviates the need
for additional dither noise. Increasing the wideband rms noise
level beyond approximately one LSB will proportionally reduce
the SNR and result in no additional improvement.

AIN

ADC

NOISE
GENERATOR

Nothing can be done externally to the ADC to significantly reduce

the inherent distortion caused by its front end. However, the
differential nonlinearity in the ADCs encoder transfer function
can be reduced by the proper use of dither (external noise that is
intentionally summed with the analog input signal).

AIN

INPUT
+

ADC

1/2 LSB RMS

Spurious-free dynamic range (SFDR) is the ratio of the rms

signal amplitude to the rms value of the peak spurious spectral
component. Two fundamental limitations to maximizing SFDR
in a high-speed ADC are the distortion produced by the front-end
amplifier and the sample-and-hold circuit; and that produced by
nonlinearity in the transfer function of the encoder portion of
the ADC. The key to achieving high SFDR is to minimize both
sources of nonlinearity.

AVCC

(b)
LARGE AMPLITUDE

DIGITAL ERROR CORRECTION LOGIC

DMID

OVR

DRY

D13
(MSB)

D12

D11

D10

D9

D8

D7

D6

D5

D4

D3

D2

D1

D0
(LSB)

Figure 7. AD6645 14-bit, 105 MSPS ADC simplified block diagram.

Analog Dialogue 40-02, February (2006)

Figure 8 shows a greatly exaggerated representation of these

nonlinearities.

The first plot shown in Figure 9 shows the undithered DNL over
a small portion of the input signal range, including two of the
subranging points, which are spaced 68.75 mV (512 LSBs) apart.
The second plot shows the DNL after adding (and later filtering
out) 155 LSBs of rms dither. This amount of dither corresponds
to approximately 20.6 dBm. Note the dramatic improvement in
the DNL.

FULL-SCALE = 2.2V p-p

25 = 32 ADC1 TRANSITIONS

OUTPUT CODE

higher levels of noise. Two ADC1 transitions cover 1024 LSBs

peak-to-peak, or approximately 155 (= 1024/6.6) LSBs rms.

68.75mV
29 =
512 LSBs

Dither noise can be generated in a number of ways. For example,

noise diodes can be used, but simply amplifying the input voltage
noise of a wideband bipolar op amp provides a more economical
solution. This approach, described in detail elsewhere (Further
Reading 3, 4, and 5) will not be discussed here.

ANALOG INPUT

Figure 8. AD6645 subranging point DNL errors (exaggerated).

With an analog input up to about 200 MHz, the distortion
components produced by the front end of the AD6645 are
negligible compared to those produced by the encoder. That is,
the static nonlinearity of the AD6645 transfer function is the chief
limitation to SFDR.

The dramatic improvement in SFDR obtainable with out-of-band

dither is shown in Figure 10, using a deep (1,048,576-point) FFT,
where the AD6645 is sampling a 35-dBm, 30.5-MHz signal at
80 MSPS. Note that the SFDR without dither is approximately
92 dBFS, compared to 108 dBFS with dithera substantial
16-dB improvement!

The goal is to select the proper amount of out-of-band dither so

that the effects of these small DNL errors are randomized across
the ADC input range, thereby reducing the average DNL error.
Experimentally, it was determined that making the peak-to-peak
dither noise cover about two ADC1 transitions gives the best
improvement in DNL. The DNL is not significantly improved with

The AD6645 ADC, introduced by Analog Devices in 2000, has

until recently represented the ultimate in SFDR performance.
In the few years since its introduction, improvements in both
process technology and circuit design have resulted in even
higher performance ADCs, such as the AD9444 (14 bits at
80 MSPS) 3 , AD9445 (14 bits at 105 MSPS/125 MSPS) 4 ,

UNDITHERED

1.5

155 LSBs RMS DITHER

1.5

512 LSBs

512 LSBs
1.0
DNL (LSBs)

DNL (LSBs)

1.0

0.5

0.5

0.5

0.5
OUTPUT CODE

OUTPUT CODE

Figure 9. AD6645 DNL plot, without and with dither.

NO DITHER

0
10
20

20
30

40

40

SFDR =
92dBFS

SFDR =
108dBFS

50
dBFS

50
dBFS

SAMPLING RATE = 80MSPS

INPUT = 30.5MHz @ 35dBm
WITH DITHER @ 20.6dBm

10

30

60
70
80

60

DITHER SIGNAL

70
80

90

90

100

100

110

120
0

10

15
20
25
FREQUENCY (MHz)

30

35

40

130

110

120
130

WITH DITHER

SAMPLING RATE = 80MSPS

INPUT = 30.5MHz @ 35dBm
NO DITHER

1,048,576-POINT FFTs,
PROCESS GAIN = 60dB

10

15
20
25
FREQUENCY (MHz)

30

35

40

Figure 10. FFT plots showing AD6645 SFDR, without and with the use of dither.

Analog Dialogue 40-02, February (2006)

NO DITHER

50mV RMS DITHER

25

25
SFDR = 125dBFS

50

50

75

75
5

100

dBFS

dBFS

SFDR = 100dBFS

100

125

125

150

150

175

175

200

10

15
20
25
FREQUENCY (MHz)

30

35

40

200

10

15
20
25
FREQUENCY (MHz)

30

35

40

1,048,576-POINT FFTs, AVERAGE OF 5 RUNS,

DATA GENERATED USING ADIsimADC AND AD9444 MODEL

Figure 11. AD9444, a 14-bit, 80-MSPS ADC; fS = 80 MSPS, fIN = 30.5 MHz, signal amplitude = 40 dBFS.
and the AD9446 (16 bits at 80 MSPS/100 MSPS) 5. These
ADCs have very high SFDR (typically greater than 90 dBc
for a 70-MHz, full-scale input signal) and low DNL. Still, the
addition of an appropriate out-of-band dither signal can improve
the SFDR under certain input signal conditions.

In certain high speed ADC applications, the addition of the

proper amount of out-of-band noise dither can improve the DNL
of the ADC and increase its SFDR. However, the effectiveness
of dither in improving SFDR is highly dependent upon the
characteristics of the ADC being considered.
b

Figure 11 models FFT plots of the AD9444, with and without

dither. It can be seen that, under the given input conditions, the
addition of dither improves the SFDR by 25 dB. The data was taken
using the ADIsimADC program6 and the AD9444 model.

ACKNOWLEDGEMENTS

Even though the results shown in Figures 10 and 11 are fairly

dramatic, it should not be assumed that the addition of out-of-band
noise dither will always improve the SFDR of the ADC under all
conditions. We reiterate that dither will not improve the linearity
of the front-end circuits of the ADC. Even with a nearly ideal front
end, the effects of dither will be highly dependent upon both the
amplitude of the input signal and the amplitude of the dither signal
itself. For example, when signals are near the full-scale input range
of the ADC, the integral nonlinearity of the transfer function may
become the limiting factor in determining SFDR, and dither will
not help. In any event, the data sheet should be studied carefully
in some cases dithered and undithered data may be shown, along
with suggestions for the amplitude and bandwidth. Dither may
be a built-in feature of newer IF-sampling ADCs.

SUMMARY

In this discussion we have considered the input-referred noise,

common to all ADCs. In precision, low-frequency measurement
applications, effects of this noise can be reduced by digitally
averaging the ADC output data, using lower sampling rates and
additional hardware. While the resolution of the ADC can actually
be increased by this averaging process, integral-nonlinearity errors
are not reduced. Only a small amount of input-referred noise is
needed to increase the resolution by the averaging technique;
however, use of increased noise requires a larger number of samples
in the average, so a point of diminishing returns is reached.

Analog Dialogue 40-02, February (2006)

The author would like to thank Bonnie Baker of Microchip

Technology and Alain Guery of Analog Devices for their
thoughtful inputs to this article.

FURTHER READING

1. Baker, Bonnie, Sometimes, Noise Can Be Good, EDN,

February 17, 2005, p. 26.
2. Brannon, Brad, Overcoming Converter Nonlinearities with
Dither, Application Note AN-410, Analog Devices, 1995.
3. Jung, Walt, Op Amp Applications, Analog Devices, 2002,
ISBN 0-916550-26-5, p. 6.165, A Simple Wideband Noise
Generator. Also available as Op Amp Applications Handbook,
Newnes, 2005, ISBN 0-7506-7844-5, p. 568.
4. Jung, Walt, Wideband Noise Generator, Ideas for Design,
Electronic Design, October 1, 1996.
5. Kester, Walt, Add Noise Dither to Blow Out ADCs Dynamic
Range, Electronic Design, Analog Applications Supplement,
November 22, 1999, pp. 20-26.
6. Ruscak, Steve and Larry Singer, Using Histogram Techniques to
Measure A/D Converter Noise, Analog Dialogue, Vol. 29-2, 1995.

REFERENCESVALID AS OF FEBRUARY 2006

1

ADI website: www.analog.com (Search) AD7730 (Go)

ADI website: www.analog.com (Search) AD6645 (Go)
3
ADI website: www.analog.com (Search) AD9444 (Go)
4
ADI website: www.analog.com (Search) AD9445 (Go)
5
ADI website: www.analog.com (Search) AD9446 (Go)
6
ADI website: www.analog.com/ADIsimADC
2